Two research teams have developed robotic devices that can be used to retrieve objects in tight or complex places. This is a step forward in automating redundant tasks in many professional fields.
Find objects and move them around
However, the future of this type of robot seemed unclear since Walmart’s categorical decision. Indeed, the American retail giant has announced it is parting with the robot fleet that supports its employees. One of her main tasks was to clear away misplaced products. However, according to Walmart, this alternative saved neither time nor money, and the human solution proved to be more effective. However, it could be that the advancement of research in this area will encourage the arrival of robots for shelves, for example.
In the same category
The CNIL fines Carrefour in excess of 3 million euros
Researchers at the University of Berkeley in California have therefore developed a system called Lateral Access maXimal Reduction of Occupancy Support Area (LAX-RAY) that can even predict the position of an object in lateral environments such as shelves if only a tiny part of it is visible is. To do this, they included three different methods of searching the system and then generating 800 random radiation environments. Then they provided LAX-RAY with a fetch robot (especially for warehouses) and an integrated camera on a real shelf.
With an efficiency of 87.3%, the system was able to determine the position of objects so precisely that the robot could push them. In the future, researchers want to investigate more complex depth models and, above all, develop pulling actions with suction cups to lift or remove the identified objects.
Multiple use to replace redundant tasks
Google researchers have developed a device called Contact-Aware Online COntext Inference (COCOI). Based on machine learning, the dynamic properties of physical objects should be integrated into a user-friendly framework in order to move them. With a simulation, but also a virtual robot, COCOI can push objects to a certain place without dropping them. Now, if the system has been shown to be very effective, scientists want to use it for things that are more difficult to use, such as: B. Clothing.
Robots that can search for objects in environments such as shelves, cabinets, and cupboards have many uses. At the moment, however, the technical challenge posed by such devices remains significant, but these two works pave the way for the use of systems that can store shelves or even retrieve objects from commercial shelves.